Authors: Nicholas Carr
Subsequent studies confirmed the existence of short-term and long-term forms of memory and provided further evidence of the importance of the consolidation phase during which the former are turned into the latter. In the 1960s, University of Pennsylvania neurologist Louis Flexner made a particularly intriguing discovery. After injecting mice with an antibiotic drug that prevented their cells from producing proteins, he found that the animals were unable to form long-term memories (about how to avoid receiving a shock while in a maze) but could continue to store short-term ones. The implication was clear: long-term memories are not just stronger forms of short-term memories. The two types of memory entail different biological processes. Storing long-term memories requires the synthesis of new proteins. Storing short-term memories does not.
Inspired by the groundbreaking results of his earlier
experiments, Kandel recruited a team of talented researchers, including physiological psychologists and cell biologists, to help him plumb the physical workings of both short-term and long-term memory. They began to meticulously trace the course of a sea slug’s neuronal signals, “one cell at a time,” as the animal learned to adapt to outside stimuli such as pokes and shocks to its body.
They quickly confirmed what Ebbinghaus had observed: the more times an experience is repeated, the longer the memory of the experience lasts. Repetition encourages consolidation. When they examined the physiological effects of repetition on individual neurons and synapses, they discovered something amazing. Not only did the concentration of neurotransmitters in synapses change, altering the strength of the existing connections between neurons, but the neurons grew entirely new synaptic terminals. The formation of long-term memories, in other words, involves not only biochemical changes but anatomical ones. That explained, Kandel realized, why memory consolidation requires new proteins. Proteins play an essential role in producing structural changes in cells.
The anatomical alterations in the slug’s relatively simple memory circuits were extensive. In one case, the researchers found that, before a long-term memory was consolidated, a particular sensory neuron had some thirteen hundred synaptic connections to about twenty-five other neurons. Only about forty percent of those connections were active—in other words, sending signals through the production of neurotransmitters. After the long-term memory had been formed, the number of synaptic connections had more than doubled, to about twenty-seven hundred, and the proportion that were active had increased from forty percent to sixty percent. The new synapses remained in place as long as the memory persisted. When the memory was allowed to fade—by discontinuing the repetition of the experience—the number of synapses eventually dropped to about fifteen hundred. The fact that, even after a memory is forgotten, the number of synapses remains a bit higher than it had been originally helps explain why it’s easier to learn something a second time.
Through the new round of
experiments, Kandel wrote in his 2006 memoir
In Search of Memory
, “we could see for the first time that the number of synapses in the brain is not fixed—it changes with learning! Moreover, long-term memory persists for as long as the anatomical changes are maintained.” The research also revealed the basic physiological difference between the two types of memory: “Short-term memory produces a change in the function of the synapse, strengthening or weakening preexisting connections; long-term memory requires anatomical changes.”
Kandel’s findings fit seamlessly with the discoveries being made about neuroplasticity by Michael Merzenich and others. Further experiments soon made it clear that the biochemical and structural changes involved in memory consolidation are not limited to slugs. They also take place in the brains of other animals, including primates.
Kandel and his colleagues had unlocked some of the secrets of memory at the cellular level. Now, they wanted to go deeper—to the molecular processes within the cells. The researchers were, as Kandel later put it, “entering completely uncharted territory.”
They looked first at the molecular changes that occur in synapses as short-term memories are formed. They found that the process involves much more than just the transmission of a neurotransmitter—glutamate, in this case—from one neuron to another. Other types of cells, called interneurons, are also involved. The interneurons produce the neurotransmitter serotonin, which fine-tunes the synaptic connection, modulating the amount of glutamate released into the synapse. Working with the biochemists James Schwartz and Paul Greengard, Kandel discovered that the fine-tuning occurs through a series of molecular signals. The serotonin released by the interneuron binds to a receptor on the membrane of the presynaptic neuron—the neuron carrying the electric pulse—which starts a chemical reaction that leads the neuron to produce a molecule called cyclic AMP. The cyclic AMP in turn activates a protein called kinase A, a catalytic enzyme that spurs the cell to release more glutamate into the synapse, thereby strengthening the synaptic connection, prolonging the electrical activity in the linked neurons, and enabling the brain to maintain the short-term memory for seconds or minutes.
The next challenge facing Kandel was to figure out how such briefly held short-term memories could be transformed into much more permanent long-term memories. What was the molecular basis of the consolidation process? Answering that question would require him to enter the realm of genetics.
In 1983, the prestigious and well-financed Howard Hughes Medical Institute asked Kandel, together with Schwartz and the Columbia University neuroscientist Richard Axel, to head a research group in molecular cognition, based at Columbia. The group soon succeeded in harvesting neurons from larval
and using them to grow, as a tissue culture in the laboratory, a basic neural circuit incorporating a presynaptic neuron, a postsynaptic neuron, and the synapse between them. To mimic the action of the modulating interneurons, the scientists injected serotonin into the culture. A single squirt of serotonin, replicating a single learning experience, triggered, as expected, a release of glutamate—producing the brief strengthening of the synapse that is characteristic of short-term memory. Five separate squirts of serotonin, in contrast, strengthened the existing synapse for days and also spurred the formation of new synaptic terminals—changes characteristic of long-term memory.
What happens after repeated injections of serotonin is that the enzyme kinase A, along with another enzyme, called MAP, moves from the neuron’s outer cytoplasm into its nucleus. There, kinase A activates a protein called CREB-1, which in turn switches on a set of genes that synthesize the proteins the neuron needs to grow new synaptic terminals. At the same time, MAP activates another protein, CREB-2, which switches off a set of genes that inhibit the growth of new terminals. Through a complex chemical process of cellular “marking,” the resulting synaptic changes are concentrated at particular regions on the surface of the neuron and perpetuated over long periods of time. It is through this elaborate process, involving extensive chemical and genetic signals and changes, that synapses become able to hold memories over the course of days or even years. “The growth and maintenance of new synaptic terminals,” writes Kandel, “makes memory persist.”
The process also says something important about how, thanks to the plasticity of our brains, our experiences continually shape our behavior and identity: “The fact that a gene must be switched on to form long-term memory shows clearly that genes are not simply determinants of behavior but are also responsive to environmental stimulation, such as learning.”
THE MENTAL LIFE
of a sea slug, it seems safe to say, is not particularly exciting. The memory circuits that Kandel and his team studied were simple ones. They involved the storage of what psychologists call “implicit” memories—the unconscious memories of past experiences that are recalled automatically in carrying out a reflexive action or rehearsing a learned skill. A slug calls on implicit memories when retracting its gill. A person draws on them when dribbling a basketball or riding a bike. As Kandel explains, an implicit memory “is recalled directly through performance, without any conscious effort or even awareness that we are drawing on memory.”
When we talk about our memories, what we’re usually referring to are the “explicit” ones—the recollections of people, events, facts, ideas, feelings, and impressions that we’re able to summon into the working memory of our conscious mind. Explicit memory encompasses everything that we say we “remember” about the past. Kandel refers to explicit memory as “complex memory”—and for good reason. The long-term storage of explicit memories involves all the biochemical and molecular processes of “synaptic consolidation” that play out in storing implicit memories. But it also requires a second form of consolidation, called “system consolidation,” which involves concerted interactions among far-flung areas of the brain. Scientists have only recently begun to document the workings of system consolidation, and many of their findings remain tentative. What’s clear, though, is that the consolidation of explicit memories involves a long and involved “conversation” between the cerebral cortex and the hippocampus.
A small, ancient part of the brain, the hippocampus lies beneath the cortex, folded deep within the medial temporal lobes. As well as being the seat of our navigational sense—it’s where London cabbies store their mental maps of the city’s roads—the hippocampus plays an important role in the formation and management of explicit memories. Much of the credit for the discovery of the hippocampus’s connection with memory storage lies with an unfortunate man named Henry Molaison. Born in 1926, Molaison was stricken with epilepsy after suffering a severe head injury in his youth. During his adult years, he experienced increasingly debilitating grand mal seizures. The source of his affliction was eventually traced to the area of his hippocampus, and in 1953 doctors removed most of the hippocampus as well as other parts of the medial temporal lobes. The surgery cured Molaison’s epilepsy, but it had an extraordinarily strange effect on his memory. His implicit memories remained intact, as did his older explicit memories. He could remember the events of his childhood in great detail. But many of his more recent explicit memories—some dating back years before the surgery—had vanished. And he was no longer able to store new explicit memories. Events slipped from his mind moments after they happened.
Molaison’s experience, meticulously documented by the English psychologist Brenda Milner, suggested that the hippocampus is essential to the consolidation of new explicit memories but that after a time many of those memories come to exist independently of the hippocampus.
Extensive experiments over the last five decades have helped untangle this conundrum. The memory of an experience seems to be stored initially not only in the cortical regions that record the experience—the auditory cortex for a memory of a sound, the visual cortex for a memory of a sight, and so forth—but also in the hippocampus. The hippocampus provides an ideal holding place for new memories because its synapses are able to change very quickly. Over the course of a few days, through a still mysterious signaling process, the hippocampus helps stabilize the memory in the cortex, beginning its transformation from a short-term memory into a long-term one. Eventually, once the memory is fully consolidated, it appears to be erased from the hippocampus. The cortex becomes its sole holding place. Fully transferring an explicit memory from the hippocampus to the cortex is a gradual process that can take many years.
That’s why so many of Molaison’s memories disappeared along with his hippocampus.
The hippocampus seems to act as something like an orchestra conductor in directing the symphony of our conscious memory. Beyond its involvement in fixing particular memories in the cortex, it is thought to play an important role in weaving together the various contemporaneous memories—visual, spatial, auditory, tactile, emotional—that are stored separately in the brain but that coalesce to form a single, seamless recollection of an event. Neuroscientists also theorize that the hippocampus helps link new memories with older ones, forming the rich mesh of neuronal connections that give memory its flexibility and depth. Many of the connections between memories are likely forged when we’re asleep and the hippocampus is relieved of some of its other cognitive chores. As the psychiatrist Daniel Siegel explains in his book
The Developing Mind
, “Though filled with a combination of seemingly random activations, aspects of the day’s experiences, and elements from the distant past, dreams may be a fundamental way in which the mind consolidates the myriad of explicit recollections into a coherent set of representations for permanent, consolidated memory.”
When our sleep suffers, studies show, so, too, does our memory.
Much remains to be learned about the workings of explicit and even implicit memory, and much of what we now know will be revised and refined through future research. But the growing body of evidence makes clear that the memory inside our heads is the product of an extraordinarily complex natural process that is, at every instant, exquisitely tuned to the unique environment in which each of us lives and the unique pattern of experiences that each of us goes through. The old botanical metaphors for memory, with their emphasis on continual, indeterminate organic growth, are, it turns out, remarkably apt. In fact, they seem to be more fitting than our new, fashionably high-tech metaphors, which equate biological memory with the precisely defined bits of digital data stored in databases and processed by computer chips. Governed by highly variable biological signals, chemical, electrical, and genetic, every aspect of human memory—the way it’s formed, maintained, connected, recalled—has almost infinite gradations. Computer memory exists as simple binary bits—ones and zeros—that are processed through fixed circuits, which can be either open or closed but nothing in between.